Re: [Xsb-development] Multi-threading support in XSB CVS and POSIX threads

[Terrance S. <ts...@cs...>]

Hi Paulo:

I'm happy to work with a standard if that is feasible.  We'll
probably be releasing version 3.0 very soon, but that doesnt mean
that we wont be able to change the thread API (at least in my
mind).  Here are some preliminary thoughts.

-- Platforms vary in their support for Posix (some dont have
   functions like thread_yeild, for instance), and of course
   Windows has a completely different mechanism.  Differences in
   implementations make a difference in how much of Posix you
   can (or want to) support.

-- XSB's threading has some features that other Prologs may not
   have.  We allow both private and dynamic code to be either
   thread-shared or thread-private, and have different mechanisms
   for allowing thread-shared tables to be evaluated, depending
   on our tabling strategy.  I dont think that this affects the
   semantics of the thread API, thought.

-- In xsb_thread_create/3 we allow users to specify the size of
   the WAM stacks with which a thread is initialized, while in
   the Posix thread create function a thread can be initialized
   with different sizes for a C stack.  These are analogous, but
   certainly different, which is one reason for attaching
   the "xsb_" to the beginning of the thread create.  Of course
   we could prepend "prolog_".

--- I'll be curious to see how your draft handles error balls
    thrown by threads, and Prolog thread cancellation.  To what
    extend is the handling of Prolog thread cancellation

Can you send me a copy of your draft API?



On Mon, 22 May 2006, Paulo Moura wrote:

>
> On 2006/05/21, at 23:50, Terrance Swift wrote:
>
> > Yes.
>
> Of course. Silly me. I was looking to the wrong user manual (I've
> both 2.71. and CVS versions on my laptop). Sorry for the wasted bits.
>
> I'm currently working with a PhD student and some Prolog developers
> on a draft standard for multi-threading programming in Prolog. The
> proposed API is based on POSIX threads and its implemented from some
> time in two Prolog compilers (SWI-Prolog and YAP). A third compiler
> (Qu-Prolog) is being updated to provide the same interface (with some
> functionality of the old QP version being moved the the draft
> proposal). Currently, the major difference between the predicates on
> the XSB CVS version and our draft is the "xsb_" prefix in the names
> of the predicates. I believe that there is an opportunity now to act
> proactively as a community on a standard multi-threading programming
> API for Prolog. We would like your support for this initiative. Let
> us know...
>
> Best regards,
>
> Paulo
>
>
> -----------------------------------------------------------------
> Paulo Jorge Lopes de Moura
> Dep. of Computer Science, University of Beira Interior
> 6201-001 Covilh=E3, Portugal
>
> Office 4.3  Ext. 3257
> Phone: +351 275319891 Fax: +351 275319899
> Email: <mailto:pm...@di...>
>
> Home page: <http://www.di.ubi.pt/~pmoura>
> Research: <http://logtalk.org/>
> -----------------------------------------------------------------
>
>
>
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job ea=
sier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronim=
o
> http://sel.as-us.falkag.net/sel?cmd______________________________________=
_________
> Xsb-development mailing list
> Xsb...@li...
> https://lists.sourceforge.net/lists/listinfo/xsb-development
>
>